The role of the audit committee and the informativeness of accounting earnings in East Asia

(1)

The role of the audit committee and the informativeness of

accounting earnings in East Asia

Tracie Woidtke

a

, Yin-Hua Yeh

b,

⁎

a

Department of Finance, Corporate Governance Center, University of Tennessee, United States b

Graduate Institute of Finance, National Chiao Tung University, 1001 Ta-Hsueh Rd., Hsinchu City, Taiwan 30010

a r t i c l e i n f o

a b s t r a c t

Article history: Received 12 March 2012 Accepted 18 December 2012 Available online 29 December 2012

Policy makers around the world have focused on corporate gover-nance reform since the Asianfinancial crisis and scandals in the United States such as the Enron debacle. In particular, policy makers have focused on the establishment of independent audit committees to improve investor confidence in reported accounting information. In a sample of East Asian companies, wefind that the negative relation between concentrated control and earnings informativeness that was documented prior to the Asianfinancial crisis persists in a more recent period, even though many corporate governance reforms have been adopted since the crisis to improvefinancial disclosure. We do, how-ever,find that earnings informativeness is strengthened by both fully independent audit committees and audit committees with a majority of independent directors with accountingfinancial or legal expertise. In addition, the increased reliability that is associated with these audit committee characteristics appears to more than offset the detrimental effect that is associated with concentrated control. The results in this paper suggest that an emphasis on audit committee independence alone may not be enough to enhance earnings informativeness. Instead, focusing on both complete independence and thefinancial or legal expertise of independent directors who are appointed to the audit committee may be a more fruitful way to increase investor confidence in accounting information, especially when ownership is concentrated. © 2012 Elsevier B.V. All rights reserved.

Jel classiﬁcation: G34 G32 Keywords: Corporate governance Ownership Earnings informativeness Audit committee 1. Introduction

The prevalence of concentrated ownership of East Asian companies (Claessens et al., 2000) has led to the belief that controlling shareholders have opportunistic incentives to take advantage of weak domestic

⁎ Corresponding author. Tel.: +886 3 513 1294.

Contents lists available atSciVerse ScienceDirect

Paci

ﬁc-Basin Finance Journal

(2)

legal systems and ineffective corporate governance mechanisms to increase their own wealth at the expense of minority shareholders (Shleifer and Vishny, 1997; La Porta et al., 1999; Johnson et al., 2000). The tactical use of pyramidal and cross-holding ownership structures exacerbates the problem because it results in a deviation of cashﬂow rights from voting rights.1_{In these companies, controlling shareholders}

are able to take control in excess of what they would have without complicated ownership structures. Thus, accounting reports that are overseen by controlling shareholders may be viewed as less credible by other investors, even when a country's accounting standards are viewed as being strong.

Ball et al. (2003)argue that accounting standards are generally viewed as high-quality in their sample of East Asian countries. In fact, Transparency ratings, which are based largely on accounting standards, rate Singapore higher than the U.S. and U.K. However, theyfind evidence consistent with lower quality of reported earnings in East Asia. In a similar vein,Leuz et al. (2003)classify three of the East Asian countries that Ball et al. study as outsider economies similar to the U.K. and U.S. because of the common law influence in these countries, but they find these countries have by far the worst earnings management ratings.Fan and Wong (2002)alsofind that concentrated voting rights and greater disparity between cash flow rights and voting rights in East Asia are associated with poor informativeness of earnings.

The potential impact of poor corporate governance structures can be far reaching. For example, research indicates that poor corporate governance is partly to blame for the Asianﬁnancial crisis (e.g.,Johnson et al., 2000). As a result of the Asianﬁnancial crisis, in particular, greater emphasis has been given to strength-ening the independence and oversight role of the board of directors of East Asian companies. The estab-lishment of audit committees on the boards, and particularly the independence of the audit committees, has been high on the agenda of policy makers in hopes of reducing information asymmetry between controlling shareholders and other investors. It is therefore of interest to academics, practitioners, and policy makers to understand how audit committees are related to the quality of accounting information from the perspective of investors.

Most of the studies related to the quality of earnings in East Asia are based on sample periods that end in thefirst half of the 1990s. That is before many East Asian countries adopted current policies on audit committees. Thus, a natural question is whether the negative relation between concentrated control and earnings informativeness persists in East Asia during a later period when mostfirms have established audit committees. Moreover, does audit committee independence increase earnings informativeness? Or are other characteristics, which are now prescribed in the U.S., but not required in East Asia, more likely to enhance investor confidence in the credibility of earnings, and do these relations depend on a firm's ownership structure? Several studies of U.S.firms indicate that the quality of earnings or integrity of financial reporting is greater when an independent director with financial expertise serves on a firm's audit committee (e.g.,Agrawal and Chadha, 2005; Karamanou and Vafeas, 2005; DeFond et al., 2005; Xie et al., 2003; Davidson et al., 2004; Krishnan, 2005) or when the audit committee is fully independent (Anderson et al., 2004). Do we observe similar audit committee characteristics (i.e., voluntary appointment of independent directors withfinancial expertise or voluntary full independence), and are these character-istics related to earnings informativeness, in East Asia?

This paper explores the relation between the informativeness of accounting earnings, measured as the relation between earnings and cumulative abnormal stock returns, and both ownership structure and audit committee characteristics. We examine 450 companies from Hong Kong, Singapore, and Malaysia. First, consistent withFan and Wong (2002), we_{find some evidence that earnings informativeness is lower when} controlling shareholders hold more voting rights. However, the convergence between the cashflow rights and voting rights of controlling shareholders is not significantly related to earnings informativeness, which differs fromFan and Wong (2002). Thus, there is some support that Fan and Wong's information effect, or desire to prevent information leakage, persists but their entrenchment effect does not for a sample in which a majority offirms have established audit committees. Second, we find that both complete independence and the proportion of independent directors with“accounting financial” or legal expertise on the audit committee are positively related to earnings informativeness. The results suggest that both limiting CEO influence, i.e., full independence, and expertise are more strongly related to earnings informativeness than

1

(La Porta et al., 1999; Claessens et al., 2000; Faccio and Lang, 2002)ﬁnd that controlling shareholders of publicly traded ﬁrms in

most countries typically have signiﬁcant control in excess of their cash ﬂow investment by using pyramidal and cross-holding ownership structures.

(3)

majority independence. Moreover, the positive effect on earnings informativeness that is associated with a higher proportion of independentfinancial or legal expertise on the audit committee appears to offset any negative effects on earning informativeness that is associated with a high level of voting rights in the hands of controlling shareholders. Finally, the results appear to be driven by the sub-sample offirms with high ownership concentration (i.e., cashflow rights). The results are not significant for the sub-sample of firms with low ownership concentration. This result is consistent with the results inKrishnan and Visvanathan (2009)andDeFond et al. (2005), which indicate the role of the audit committee composition is a com-plement to afirm's governance structure rather than a substitute. In particular, when the incentive effects of cashflow ownership are high, higher quality audit committees are viewed as being more effective.2

This paper contributes to both the literature on corporate governance and the literature on accounting quality. Thefindings indicate that complete audit committee independence and independent directors who have legal or accountingfinancial expertise improve the general accounting quality of a firm. In particular, when controlling shareholders have a high level of cashflow rights and the audit committee is fully independent or has a higher proportion of independent directors with accountingfinancial expertise, earning informativeness is enhanced. This suggests that controlling shareholders whose incentives are more likely to be aligned with minority shareholders because they have high cashflow rights (which investors can observe), can make the financial statements of a firm more credible by establishing an independent audit committee withfinancial expertise (which investors can observe) in an environment where control rights tend to exceed cashflow rights and preventing information leakage may be desirable (the degree of which is difficult for investors to observe). In other words, it appears that firms with high cashflow ownership can appoint independent directors with certain expertise to signal that the incentive alignment effect associated with their ownership outweighs both the potential entrenchment or infor-mation effects associated with concentrated control.

The remainder of the paper is organized as follows.Section 2develops the motivation and hypotheses tested in the paper.Section 3describes the sample and country level statistics.Section 4presents the model and provides variable descriptive statistics.Section 5reports empirical results, andSection 6concludes. 2. Motivation and development of hypotheses

2.1. Ownership structure and quality of earnings

Warfield et al. (1995)investigate how the separation of corporate ownership and control affects both the informativeness of earnings and the accounting choices of managers within the U.S. Theyfind that higher managerial ownership is associated both with higher informativeness of earnings and lower discretionary accruals, supporting the incentive alignment effects of cash_{flow ownership concentration.}

Wang (2006)extends the study ofWarfield et al. (1995)andfinds higher earnings quality for U.S. family firms, again supporting the incentive alignment effects of cash flow ownership concentration. However,

Agrawal and Chadha (2005)find the probability of earnings restatements in the U.S. is higher when the CEO belongs to the founding family. And Francis et al. (2005)find no association between management ownership and earnings informativeness in a sample of U.S.firms, but instead find that the informativeness of earnings is significantly lower when cash flow rights are separated from voting rights via dual class shares.

The ownership of listed companies in most countries outside the U.S. is typically concentrated in the hands of controlling shareholders, or an ultimate owner, and disparity between cashflow rights and voting rights is not uncommon (La Porta et al., 1999).Fan and Wong (2002)present two potential effects of concentrated ownership in East Asia on earnings informativeness, the entrenchment effect and informa-tion effect. The entrenchment effect argues that as the divergence between cashflow and voting rights increases, the controlling shareholder will have a greater incentive to expropriate wealth from minority shareholders and thus predicts a negative relation between earnings informativeness and the divergence between cashflow and voting rights. The information effect argues that high managerial ownership in East

2_{Warﬁeld et al. (1995)}

document a positive relation between managerial ownership and earnings quality in the U.S. supporting the incentive alignment effects of cash ownership.Wang (2006)finds a consistent result for family firms in the U.S. He argues that U.S. familyfirms have higher earnings quality because of the incentive effects of greater cash flow ownership.

(4)

Asia may be a response to a need to operate in greater secrecy. Concentrated control directs the decision rights to a small group of people who possess specific knowledge (Jensen and Meckling, 1992; Christie et al., 1993), which prevents confidential information from leaking out and reduces the transaction cost of knowledge. Preventing the leakage of specific knowledge can be valuable in East Asia where lobbying activities are common and lucrative. Because concentrated control naturally establishes a barrier to informationflow to the populace and discourages external competition, concentrated control is associated with opaquefinancial reporting and low earnings informativeness. In their study of seven East Asian economies, Fan and Wongfind support for both the entrenchment and information effects.

We are interested in whether the negative relations that Fan and Wongﬁnd persist in a more recent time period in countries that adopted stricter corporate governance policies after the Asianﬁnancial crisis. In addition, we extend their analysis to see whether audit committee independence and composition are related to the informativess of earnings in this region.

2.2. Audit committee characteristics and quality of earnings

Afirm's audit committee primarily oversees the financial reporting process of a firm. It meets regularly with the outside auditors and internalfinancial managers of the firm to review the financial statements, audit process, and internal accounting controls of thefirm. Thus, audit committees potentially serve an important oversight role and can affect the quality of reported earnings. Supporting this view,McMullen (1996)finds that firms with well-established audit committees are more likely to have reliable financial reporting (i.e., the absence of errors, irregularities, or illegal acts). Two characteristics of audit committees appear to be of particular importance when measuring effectiveness within the U.S.: audit committee independence andfinancial sophistication.

Several studies suggest that more independent audit committees are more effective. For example,Klein (2002)documents a negative relationship between audit committee independence and abnormal accruals.

Carcello and Neal (2003)report that distressedfirms that have more independent audit committees are less likely to be issued going-concern reports by their auditors, and more independent audit committees are more effective in shielding auditors from dismissal when a new going-concern report is issued.Chen and Zhou (2007)find that more independent audit committees were quicker to dismiss Arthur Andersen as theirfirm's auditor after Andersen's credibility was threatened around the Enron scandal.Krishnan (2005)

finds that independent audit committees are significantly less likely to be associated with incidences of internal control problems. However, recent reforms in the U.S. have dictated that firms have some committees, including the audit committee, be comprised entirely of outside directors. An argument for complete independence is that it provides independent committee members an opportunity to talk outside of management's hearing, and thus, reduces management's influence over the committee. Consistent with this view, the results inShivdasani and Yermack (1999)suggest that CEOs do not exert as much control over the director selection process when they do not serve on the nominating committee. Similarly,

Anderson et al. (2004)find that fully independent audit committees are associated with a significantly lower cost of debt, suggesting more reliable_{financial reporting in these firms.}

DeFond et al. (2005)state that one of the most controversial provisions of Sarbanes-Oxley requires public companies in the U.S. to disclose whether they have afinancial expert on their audit committee. The argument is that_{financial expertise is necessary to ensure that audit committees fulfill their primary} responsibilities of overseeing thefinancial reporting process and ensuring high-quality financial reporting. Numerous studies provide support for this view. For instance,DeFond et al. (2005)find a positive stock market reaction to the appointment of an accountingfinancial expert to the audit committee. Similarly,

Davidson et al. (2004)find a significantly positive stock price reaction when new members of audit committees havefinancial expertise.Xie et al. (2003)find the financial sophistication of audit committee members appears to be an important factor in constraining the propensity of managers to engage in earnings management.Agrawal and Chadha (2005)alsofind that the probability of restatement is lower in companies whose boards or audit committees have an independent director withfinancial expertise but is not significantly related to independence alone.Karamanou and Vafeas (2005)find that the financial expertise of audit committees is associated with an increased probability of management earnings forecast updates, more informative good news forecasts, and more positive stock market reactions to management forecasts.Krishnan (2005)shows that audit committees withfinancial expertise are significantly less likely

(5)

to be associated with incidences of internal control problems. Moreover,Krishnan and Visvanathan (2009)

find that auditors charge lower fees for firms when its audit committee includes financial experts. AndChen and Zhou (2007)find audit committees with greater financial expertise were quicker to dismiss Arthur Andersen as theirfirm's auditor when Andersen's credibility was threatened around the Enron scandal. More recently,Krishnan et al. (2011)find the presence of directors with legal backgrounds on the audit committee is associated with higher financial reporting quality. Additional tests indicate a positive association between changes in legal expertise and changes infinancial reporting quality, suggesting that legal expertise serves as a monitor rather than as a signal offinancial reporting quality.

According to theAsian Corporate Governance Association (2000), the Asian Financial Crisis caused a resurgence of interest in corporate governance among government and business circles. In particular, attention was given to improving disclosure. U.S. studies generally support the view that more independent audit committees and audit committees with greater financial sophistication are more effective in monitoring the corporatefinancial accounting process. However, the mere existence of an audit committee without certain characteristics is not found to make a difference. For example,Beasley (1996)reports that the incidence offinancial statement fraud is not related to the existence of an audit committee.

Of particular interest in East Asia, where the disparity between control and cashflow ownership is difficult to observe, is that minority shareholders can observe the existence, independence, and com-position of the audit committee from annual reports. Because Fan and Wongfind that earnings are less reliable when control is concentrated, we investigate which characteristics of the audit committee might be associated with more reliable earnings, especially when ownership is concentrated. In a sample of Hong Kong listed companies,Chen and Jaggi (2000)find that the ratio of independent non-executive directors to the total number of directors on corporate boards as a whole is positively associated with the compre-hensiveness offinancial disclosure.Zain et al. (2006)conduct a survey in Malaysia andfind that internal auditors feel they are able to contribute more to the external audit when audit committees are more independent and have greaterfinancial expertise. To the extent that more comprehensive financial dis-closure is associated with more reliable earnings, we might expect the presence of independent directors withfinancial or legal expertise on the audit committee, in particular, to be positively associated with the reliability or informativeness of earnings.

Hypothesis 1. Audit committee independence and ﬁnancial sophistication strengthen earnings informativeness.

When controlling shareholders assign directors to the audit committee, they can potentially increase the credibility of the accounting earnings reports, and hence earnings informativeness, by choosing independent financial and legal experts if the objectivity and professional backgrounds of these members are seen to contribute to high-qualityfinancial statements. Because investors can observe these appointments, any negative entrenchment or information effect that is associated with concentrated control might be mitigated by an improvement in both transparency and overall accounting quality associated with these appointments. Hypothesis 2. The combined effect of audit committee independence and financial sophistication on earnings informativeness mitigates the negative impact of concentrated control on earnings informativeness. 2.3. Audit committee as a substitute or complement

La Porta et al. (2002)andClaessens et al. (2002)report that higher levels of cashﬂow rights align the interests of controlling shareholders with those of minority shareholders because controlling shareholders gain more from increasing shareholder wealth than they lose by foregoing expropriation. This interest alignment encourages controlling shareholders to run the business properly, which gives rise to the positive incentive effect. Consistent with this, the results inYeh et al. (2001)suggest that high levels of family ownership and low levels of family board representation are effective ways of mitigating the separation between cashﬂow rights and control when family control is central.

To the extent that investors can identify high cash_{ﬂow rights without excessive voting rights, they will} associate high ownership concentration with aligned incentives, and it may not be necessary to incur the extra cost of setting up an independent audit committee. However, if it is difﬁcult for investors to disentangle

(6)

cashflow rights from control rights, controlling shareholders with high levels of ownership concentration may be associated with misaligned incentives or viewed as protecting proprietary information. In this case, establishing an audit committee with independent directors with legal orfinancial expertise may be as-sociated with more reliable reported earnings and a stronger corporate governance system even though investors are unsure of the incentives of the controlling shareholders. Consistent with this view,Yeh and Woidtke (2005) find the proportion of directors represented by a controlling family appears to be a reasonable proxy for the quality of corporate governance at thefirm level when investor protection is relatively weak and it is difficult to determine the degree of separation between ownership and control. Hypothesis 3a. Companies with controlling shareholders who hold a high level of cashflow rights do not need an audit committee to strengthen earnings informativeness. The positive incentive effects associated with cashflow rights held by controlling shareholders are viewed as a substitute for an audit committee. Hypothesis 3b. Companies with controlling shareholders who hold high levels of cashflow rights can strengthen earnings informativeness through the establishment of an audit committee. The audit committee is viewed as a complement to the positive incentive effects of the cashflow rights held by controlling shareholders.

3. Sample description and country level statistics 3.1. Sample description

The sample in this paper includes the largest 150 listed companies, based on their market values from annual reports for the ﬁscal year 2000, from three East Asian countries—Hong Kong, Singapore, and Malaysia. The total sample size is 450 companies. Due to the intensity of hand collecting data, we focus on these three East Asian countries because they represent three out of four East Asian countries that are included inBall et al. (2003),Fan and Wong (2002), andLeuz et al. (2003). We are also able to collect data on audit committee composition for these countries in order to examine whether the composition of the audit committee is associated with improved earnings reliability. Moreover, characteristics common to East Asia, such as concentrated ownership and family control, are common in these countries as well. The results in the aforementioned studies suggest that even though the quantity of earnings information may be increasing and accounting standards have become stricter in these countries, the quality of earnings remains poor during their sample period. We would like to examine whether this relation persists in a more recent time period after stricter corporate governance policies that focus on building stronger boards have been adopted.3

Although all three countries adopted some reforms to strengthen the board of directors prior to 1995, they also introduced additional reforms after this time period. For example, majority independent audit committees comprised of at least three members have been compulsory in Singapore since 1989. The Singapore stock exchange sought to strengthen audit committees by adding Chapter 9B to its Listing Manual and making its provisions mandatory in 1996. Chapter 9B covered detailed workings of these committees including their establishment, membership, roles and duties. But in 1998, the exchange decided to remove the compulsory nature of these more detailed rules and transfer them to a Best Practices Guide outside the Listing Manual.

Compulsory audit committees were adopted in Malaysia in 1994. Similar to Singapore, the audit committee must have at least three members with the majority being independent. The regulatory authorities also established the High Level Finance Committee on Corporate Governance in March 1998. The committee subsequently published a“Proposed Malaysian Code on Corporate Governance.” Among the conclusions the committee reached was the need to improve the accuracy and timeliness of disclosure. In line with recommendations from the committee, the Securities Commission in Malaysia introduced expanded disclosure requirements during 1998 and 1999.

3_{The three countries in our study have either 4 or 5 provisions out of 6 in place to protect shareholder rights, compared to an} average score of 4 for common law countries and 5 for the United States (La Porta et al., 1998). In addition,Credit Lyonnais Securities

Asia, (2004)rated these three countries near the top of ten Asian emerging markets in 2004 based on the followingﬁve dimensions:

rules and regulations; enforcement; political/regulatory interference; the international“Generally Accepted Accounting Principles”; and institutional mechanisms and corporate governance culture.

(7)

The Stock Exchange of Hong Kong amended its rules in 1992 to require companies to have a minimum of two independent directors and to encourage greater disclosure, accountability, and the establishment of audit committees. Since January 1999, every listed company has been expected to set up an audit committee; however, it is not compulsory. The listing rules simply require that companies explain whether or not they are complying.4

In sum, none of the three countries required fully independent audit committees or the inclusion of an audit committee member with accountingﬁnancial expertise during our sample period. However, all three countries recommended increased independence andﬁnancial expertise or enhanced disclosure of the audit committee's independence, expertise, and operations in the annual report during our sample period. 3.2. Country level descriptive statistics

Table 1presents descriptive statistics on ownership structure and audit committee independence for each country in the sample. The data on ownership and audit committee composition are obtained from company annual reports. Panel A shows that all three countries are dominated by family control. The ultimate owner of a company is a family for 72% of the entire sample, and ranges from 64% in Malaysia to 77% in Singapore.

Panel B presents data on the establishment of audit committees. Since audit committees are required in both Singapore and Malaysia, it is not surprising tofind that 100% of the sample firms in these countries have established an audit committee. In Hong Kong, where audit committees are strongly recommended but not required, approximately 75% of the samplefirms have established an audit committee.

Panel C presents the data on audit committee independence. There are 10ﬁrms with established audit committees that provide no data on audit committee independence or composition (two in Singapore and eight in Hong Kong) and 38ﬁrms in Hong Kong with no established audit committee. Consistent with

Klein (1998), we use board composition as a proxy for audit committee composition in the absence of an audit committee since the entire board would presumably take on the role of the audit committee in this case.5

Even though audit committees are not compulsory in Hong Kong and it has the highest proportion of companies with no audit committee, Hong Kong has the highest proportion of companies with audit committees that are composed solely of independent directors (44%). In contrast, in the countries where majority independent audit committees are compulsory, less than a third of the audit committees are fully independent. Twenty-nine percent of the companies in Singapore have fully independent audit committees, and a mere 7% of companies in Malaysia have fully independent audit committees. While approximately 80% ofﬁrms in the full sample have audit committees with at least two-thirds independence, further analysis reveals the vast majority of audit committees in these countries cluster around two-thirds independence. 4. Description of the model and variables

4.1. Earnings and returns

Following the literature, we measure earnings informativeness as the correlation between stock returns and reported earnings for the correspondingﬁscal year (see, for example,Wang, 2006; Francis et al., 2005; Fan and Wong, 2002; War_{ﬁeld et al., 1995}). Thus, to measure the informativeness of earnings, conditional on ownership structure, audit committee independence, and audit committee composition, we use the following expanded regression model:

CARi¼ a0þ a1ðEarningsiÞ þ a2ðEarningsi Voting RightsiÞ þ a3ðEarningsi Cash Flow to Voting RightsiÞ

þβ Earningsð i Audit Committee CharacteristicsiÞ þ a4ðEarningsi Firm SizeiÞ þ a5ðEarningsi LeverageiÞ

þa6ðEarningsi Market−to−Book ratioiÞ þ γ Firm Level Control Variablesð Þ

þ Country−and Industry−Level Fixed Effectsð Þ þ ui

4

For more information, see the Survey of Corporate Governance by theAsian Corporate Governance Association (2000). 5

Our results are not sensitive to this treatment. We obtain similar results when we assume 0% independence in the absence of an audit committee.

(8)

FollowingFan and Wong (2002), CAR is calculated as afirm's annual return less the market annual return over the same period corresponding to the release of earnings information for afirm's 2000 fiscal year. The annual returns are thus calculated by compounding monthly returns beginning twelve months prior to the latest date by which afirm is required to disclose its 2000 annual report by law or listing rules. This generally incorporates the nine months prior to and the 3 months following a_{firm's fiscal year-end.} Earnings is the net earnings that are reported forfiscal year 2000 divided by the market value of equity at the beginning of 2000.

Table 2presents summary statistics for the full sample for returns, earnings, ownership structure, audit committee characteristics, and control variables. The stock return and accounting data are obtained from Datastream and Compustat. Panel A reports an average (median) CAR, or abnormal annual return, for the sample of 4.06% (−0.30%). We also compute CAR using two-year returns for 2000 and 2001 throughout our analysis and obtain similar results. Average (median) earnings are 5.67% (6.33%).

4.2. Ownership structure

We examine two aspects of ownership structure: the level of voting rights and the divergence between cashflow and voting rights. Voting Rights is defined as the controlling voting rights of the shareholders of the firm, and Cash Flow to Voting Rights is the ratio of cash flow to voting rights of the controlling shareholder. FollowingLa Porta et al. (1999)andClaessens et al. (2000), we consider both direct and indirect control when calculating the voting rights of the ultimate owner, or controlling shareholder, of afirm. Direct control

Table 1

Ultimate control type and audit committee independence by country.

The sample consists of the largest 150 listedfirms each from Hong Kong, Singapore, and Malaysia. Panel A identifies the largest shareholder in eachfirm by type. Panel B presents the establishment of audit committees. Panel C presents the frequency of firms according to the proportion of independent directors appointed to the audit committee.

Panel A: Ultimate control owner by type

Type Hong Kong Singapore Malaysia ALL

N Fraction (%) N Fraction (%) N Fraction (%) N Fraction (%)

Family 113 75.33 115 76.67 96 64.00 324 72.00

Government 24 16.00 29 19.33 21 14.00 74 16.44

Other 13 8.67 6 4.00 33 22.00 52 11.56

Total 150 100.00 150 100.00 150 100.00 450 100.00

Panel B: Establishment of audit committee

Audit committee Hong Kong Singapore Malaysia ALL

N Fraction (%) N Fraction (%) N Fraction (%) N Fraction (%) Yesa

112 74.67 150 100.00 150 100.00 412 91.56

Nob

38 25.33 0 0.00 0 0.00 38 8.44

Total 150 100.00 150 100.00 150 100.00 450 100.00

Panel C: Proportion of independent directors on the audit committee

Proportion of independence Hong Kong Singapore Malaysia ALL

N Fraction (%) N Fraction (%) N Fraction (%) N Fraction (%)

0.00% 18 12.00 0 0.00 0 0.00 18 4.00 0.01% to 33.32% 24 16.00 4 2.67 2 1.33 30 6.67 33.33% to 66.65% 10 6.67 6 4.00 28 18.67 44 9.78 66.66% to 99.99% 32 21.33 97 64.67 109 72.67 238 52.89 100.00% 66 44.00 43 28.67 11 7.33 120 26.67 Total 150 100.00 150 100.00 150 100.00 450 100.00 a

Eight Hong Kong and two Singapore companies indicate that they have set up an audit committee, but no further details of composition or independence are given. We therefore use board independence as a proxy for independence.

(9)

includes the voting rights through shares that are registered in the name of the controlling shareholder. Indirect control includes voting rights through shares that are held by entities that the controlling shareholder controls. For each company, we carefully trace the chains of control through the relationships for a specific group, if there is one, all the way back to the ultimate owner. The ultimate owner(s) could be a family or an individual, the state, a widely heldfinancial institution, a widely held corporation, or other as defined inLa Porta et al. (1999). After identifying this information, all information is converted into a clear group map that details the ultimate owner and interlocking stakes between firm groups. Direct voting rights are then calculated as the sum of the fraction of shares that are registered to the ultimate owner, and the indirect voting rights are calculated as the“weakest link” in the chain of shares (lowest percentage of all) held by the firms that the ultimate owner controls. A detailed example is provided inAppendix A.

Panel B ofTable 2shows that, consistent with thefindings of other studies, ownership is concentrated and there is a divergence between cashflow ownership and control. The average (median) voting rights of controlling shareholders in our sample is 46.5% (46.6%), but the average (median) cashflow rights is lower, at 32.6% (30.5%). Thus, cashflow rights only account for 70% of voting rights on average.

Table 2

Full sample descriptive statistics.

The sample consists of 450firms from Hong Kong, Singapore, and Malaysia. Panel A presents abnormal stock returns (CARs) and earnings. CAR is the cumulative 12-month market-adjusted stock return for the 12 months ending on the earliest report date for a firm's 2000 fiscal year earnings. Earnings equals reported net earnings for fiscal year 2000 divided by the market value of equity at the beginning of 2000. Panel B presents the statistics on ownership structure. Voting Rights is the sum of the direct and indirect voting rights (a detailed description is provided inAppendix A). Cash Flow-to-Voting Rights is the ratio of the cashflow rights or ownership to the voting rights of the ultimate owner. Panel C presents descriptive statistics on audit committee independence. Panel D presents the statistics on the audit committee composition. Non-accountingfinancial experts are defined as independent directors who are currently or were previously employed as executives in other publicly held corporations. Accountingfinancial experts are defined as independent directors with either experience as a Chief Financial Officer or a Certified Public Accountant. Legal experts are defined as independent directors who also practice law. Accounting financial expert (Legal expert) represented is equal to one when at least one accountingfinancial (legal) expert sits on the audit committee. Majority AF-L Expert equals one when more than 50% of the audit committee is represented by either accountingfinancial or legal experts. Panel E presents summary statistics for control variables, all measured as of the beginning of 2000. Firm size is measured as the natural log of the book value of the total assets of a firm. Leverage is defined as the ratio of the book value of debt to total assets. The market-to-book ratio is defined as the market value of equity over the book value of total assets.

Variables Average Std Dev Q1 Median Q3

A. Abnormal stock returns and earnings

CAR (%) 4.06 38.44 −20.43 −0.30 22.15

Earnings (%) 5.67 31.29 2.04 6.33 10.89

B. Ownership structure

Voting rights (%) 46.48 17.10 32.8 46.56 59.56

Cashﬂow rights (%) 32.56 18.48 19.15 30.48 43.52

Cashﬂow to voting rights 0.70 0.28 0.50 0.65 1.00

C. Audit committee independence

Number of independent directors 2.10 0.97 2.00 2.00 3.00

Proportion of independent directors 0.69 0.26 0.67 0.67 1.00

100% independence 0.27 0.44 0.00 0.00 1.00

D. Audit committee composition

Proportion of non-accountingﬁnancial experts 0.34 0.27 0.17 0.34 0.50

Proportion of accountingﬁnancial experts 0.35 0.32 0.00 0.33 0.67

Proportion of legal experts 0.09 0.19 0.00 0.00 0.00

Accountingﬁnancial expert represented 0.64 0.48 0.00 1.00 1.00

Legal director represented 0.22 0.42 0.00 0.00 0.00

Majority AF-L expert 0.40 0.49 0.00 0.00 1.00

E. Control variables

Total assets (million U.S. dollars) 2448.3 7496.1 682.5 465.1 1273.0

Leverage 50.9 33.0 28.5 46.7 70.7

(10)

4.3. Audit committee characteristics

Audit Committee Characteristics include two primary measures of interest: independence and composition. We deﬁne two measures of Audit Committee Independence to examine the effects of independence on earnings informativeness. %Audit Committee Independence is the ratio of independent directors to the total number of directors that are appointed to the audit committee; and 100% Audit Committee Independence is an indicator variable that equals one when the audit committee of a company consists entirely of independent directors, and equals zero, otherwise. We use both measures to examine whether greater independence or full independence is related to earnings informativeness. Consistent with the results inTable 1, Panel C ofTable 2

shows that only 27% ofﬁrms in the full sample have fully independent audit committees. Instead, the median ﬁrm meets the majority independence requirement by appointing two independent directors out of a total of three audit committee members.

Following existing studies (e.g.,Xie et al., 2003; Davidson et al., 2004; Agrawal and Chadha, 2005; DeFond et al., 2005), we further classify independent directors according to theirfinancial expertise. We look at both the proportion of audit committee seats that is held by each type of independent expert and whether a particular type of independent expert is represented on an audit committee. Non-accountingfinancial experts are defined as independent directors who are currently or were previously employed as executives in other publicly held corporations. %Non-accounting Financial Experts is defined as the ratio of independent non-accounting financial experts to the total number of directors on the audit committee. Accounting financial experts are defined as independent directors with experience as either a Chief Financial Officer or Certified Public Accountants.6_{%Accounting Financial Experts is de}_{fined as the ratio of independent accounting}

ﬁnancial experts to the total number of directors on the audit committee. FollowingXie et al. (2003)and

Krishnan et al. (2011), we additionally include legal expertise. Legal experts are defined as independent directors who also practice law. %Legal Experts is defined as the ratio of independent legal directors to the total number of directors on the audit committee. Finally, Majority AF-L Expert is an indicator variable that equals one when more than 50% of the directors on the audit committee of a_{firm are either accounting financial or} legal independent experts as defined above.

Panel D ofTable 2contains summary statistics on audit committee composition. Non-accounting and accountingfinancial experts are the most common types of independent experts who are appointed to audit committees. On average, 34% of independent directors on audit committees are non-accounting financial experts and 35% are accounting financial experts. In contrast, only 9% of independent directors who are appointed to audit committees are legal experts during our sample period. Moreover, 64% of audit committees appoint at least one accountingfinancial expert, but only 22% of audit committees appoint at least one legal expert. Finally, wefind that 40% of sample firms appoint a majority of independent directors with either accountingfinancial or legal expertise to their audit committees.

4.4. Control variables

We additionally include a set of control variables that is similar to that used in Fan and Wong to control for observed variations in the earnings–return relation that result from causes other than ownership structure and audit committee independence or composition. All values are taken as of the beginning of 2000, i.e., atfiscal year-end 1999 for accounting data. Firm Size is measured as the natural log of the book value of total assets. Leverage is defined as the ratio of the book value of debt to total assets. The Market-to-Book ratio is defined as the market value of equity over the book value of assets. Panel E ofTable 2reports summary statistics for these variables. Average (median) book value of total assets in the sample is 2,448.3 (465.1) million U.S. dollars. Average leverage is 50.9%. Finally, the average market-to-book ratio in the sample is 1.31.

6

Generally speaking, companies that list regulations or best practices require at least one independent director withfinancial or accounting expertise, but the definition of expertise is fairly loose. For example, work experience in the financial or accounting department of a corporation could qualify directors as havingfinancial or accounting expertise. However, this information is typically not disclosed for directors who are listed as employees of other companies in annual reports. We therefore do not classify these directors as accountingfinancial experts but instead classify them as non-accounting financial experts. To the extent that these directors are as qualified to certify the informativeness of earnings as are accounting financial experts (classified as such based on our stricter definition), we should find similar results for both non-accounting and accounting financial experts.

(11)

5. Empirical results

We include indicator variables to control for industry-levelfixed effects in all of our regressions and country-levelfixed effects when appropriate. Industries are classified according to the methodology in

Claessens et al. (2000). The country andfixed effects are not reported in the tables for the sake of brevity. Thefirst set of empirical results examines the basic relations between stock returns and earnings in our sample. The results are reported inTable 3. Column 1 presents results for the full sample, and Columns 2 through 4 present results for each country individually. Similar toFan and Wong (2002), the coefficient for earnings for the full sample has a positive and significant coefficient suggesting that earnings have an information role in East Asia. The coef_{ficient is also positive for each country, though it is not significant for} Singapore.

5.1. Earnings informativeness and audit committee independence

We next test the informativeness of earnings conditional on ownership structure and audit committee independence. The empirical results are presented inTable 4. Columns 1 through 3 present results for the full sample. For comparison to previous studies, Column 1 follows the speciﬁcation used in existing studies by examining only the impact of variables on the earnings–return relation, i.e., includes only variables interacted with earnings. However, the remainder of the analyses in this paper includes non-earnings variables separately to control for any impact they may have on returns separate from reported earnings.7

Given the non-compulsory establishment of audit committees but higher voluntary compliance for fully independent audit committees, we further analyze Hong Kong separately from Malaysia and Singapore. Columns 4 and 5 present results for these separate country sub-samples.

Consistent withFan and Wong (2002), wefind a negative coefficient for earnings conditional on voting rights for the full sample. In unreported tests, the relation for earnings conditional on voting rights remains significant when we simply add earnings conditional on the two measures of audit committee inde-pendence to the specification in Column 1. However, we find its significance level is affected by the specification. Its significance level drops below conventional levels when both non-earnings variables and earnings conditional on 100% Audit Committee Independence are included in the specification in Column 3. Thus, wefind the negative relation between earnings informativeness and voting rights found in previous studies persists during our sample period when we follow similar specifications. However, the relation weakens somewhat when we add control variables not included in these studies. We additionallyfind a positive coefficient for earnings conditional on cash flow to voting rights, but unlike previous studies, it is not significant in any of the specifications.

Table 3

Simple regressions of stock returns on earnings.

This table presents a regression analysis of earnings informativeness as measured by the earnings–return relation. The dependent variable, CAR, is the cumulative 12-month market-adjusted stock return for the 12 months ending on the earliest report date for a firm's 2000 fiscal year earnings. Earnings equals reported net earnings for fiscal year 2000 divided by the market value of equity at the beginning of 2000. All regressions include industryfixed effects and country fixed effects when appropriate (not reported). T-values are reported in parentheses. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.

Independent variables Dependent variable: CAR

Full sample Hong Kong Singapore Malaysia

Industryﬁxed effects Yes Yes Yes Yes

Countryﬁxed effects Yes No No No

Intercept 2.81 (1.55) 16.32 (3.98)⁎⁎⁎ −3.01 (−1.15) −5.02 (−2.76)⁎⁎⁎

Earnings 0.23 (4.07)⁎⁎⁎ 0.41 (3.45)⁎⁎⁎ 0.06 (0.94) 0.39 (3.82)⁎⁎⁎

N 450 150 150 150

R2

0.04 0.07 0.01 0.09

7_{We also perform our entire analysis using only interaction terms similar to the speciﬁcation in Column 1. Our audit committee} results are robust to the different speciﬁcations.

(12)

Table 4

Audit committee independence and earnings informativeness.

This table presents a regression analysis of the relation between audit committee independence and earnings informativeness as measured by the earnings–return relation. The dependent variable,

CAR, is the cumulative 12-month market-adjusted stock return for the 12 months ending on the earliest report date for aﬁrm's 2000 ﬁscal year earnings. Earnings equals reported net earnings for

ﬁscal year 2000 divided by the market value of equity at the beginning of 2000. Voting Rights is the sum of the direct and indirect voting rights (a detailed description is provided inAppendix A).

Cash Flow-to-Voting Rights is the ratio of the cashﬂow rights or ownership to the voting rights of the ultimate owner. %Audit Committee Independence is the number of independent directors over the total number of directors on the audit committee when an audit committee exists, and is the number of independent directors over the total number of directors when an audit committee does not exist. The 100% Audit Committee Independence variable equals one when the audit committee of a company consists entirely of independent directors. Size is the natural logarithm of the book value of assets in millions of U.S. dollars at the beginning of 2000. Market-to-Book is the market value of equity divided by the book value of total assets at the beginning of 2000. Leverage is the

total liabilities divided by the total assets at the beginning of 2000. All regressions include industryﬁxed effects and country ﬁxed effects when appropriate (not reported). T-values are reported in

parentheses. ***, **, and * represent signiﬁcance at the 1%, 5%, and 10% levels, respectively.

Full sample Hong Kong Malaysia and Singapore

Industryﬁxed effects Yes Yes Yes Yes Yes

Countryﬁxed effects Yes Yes Yes No Yes

Intercept −17.59 (−3.39)⁎⁎⁎ −39.19 (−2.09)⁎⁎ −33.51 (−1.84)⁎ −17.92 (−0.42) −23.79 (−1.26)

Earnings 1.38 (2.65)⁎⁎⁎ 0.11 (0.13) 1.00 (1.74)⁎ 0.92 (0.58) −0.99 (−0.63)

Earnings*Voting Rights −0.86 (−2.54)⁎⁎⁎ −0.64 (−1.71)⁎ −0.51 (−1.22) −0.93 (−0.85) −0.05 (−0.09)

Earnings*Cash Flow-to-Voting Rights 0.28 (0.96) 0.48 (1.43) 0.26 (0.85) 1.10 (1.43) 0.08 (0.23)

Earnings*%Audit Committee Independence 0.56 (1.76)⁎

Earnings*100% Audit Committee Independence 0.28 (1.97)⁎⁎ 0.62 (2.25)⁎⁎ 0.45 (1.43)

Earnings*Size −0.06 (−1.49) −0.02 (−0.34) −0.05 (−1.39) −0.08 (−0.84) 0.08 (0.78)

Earnings*Leverage −0.00 (−0.01) 0.04 (0.11) 0.12 (0.36) −0.07 (−0.11) 0.03 (0.06)

Earnings*Market-to-Book 0.30 (2.00)⁎⁎ 0.29 (1.96)⁎ 0.29 (1.93)⁎ 0.34 (1.14) 0.54 (2.40)⁎⁎

Voting rights 0.05 (0.45) 0.05 (0.49) 0.24 (0.98) −0.01 (−0.10)

Cash Flow-to-Voting Rights 1.36 (0.22) 1.83 (0.29) −1.53 (−0.09) 1.33 (0.23)

%Audit Committee Independence 8.79 (1.28)

100% Audit Committee Independence Dummy 7.26 (1.55) 7.94 (0.91) 10.4 (1.27)

Size 0.94 (0.75) 0.94 (0.76) 0.25 (0.09) 1.74 (1.37) Leverage 3.35 (0.6) 2.47 (0.44) 13.74 (1.16) −8.85 (−1.51) Market-to-Book −0.99 (−2.02)⁎⁎ −0.95 (−1.94)⁎ −0.78 (−1.13) −0.62 (−0.29) R2 0.22 0.23 0.23 0.29 0.15 N 450 450 450 150 300 T. Woidtke, Y.-H. Yeh / Paci ﬁ c-Basin Finance Journal 23 (2013) 1– 24

(13)

The positive coefficient for Earnings * Audit Committee Independence suggests that earnings are more informative with more independent audit committees. In unreported tests, we substitute an indicator variable which equals 1 when at least 67%, the median value, of audit committee members are independent. The coefficient for this interaction is not significant suggesting that 100% independence is driving the positive relation, not having at least a two-thirds majority. Consistent with this interpretation, %Audit Committee Independence is significant for the Hong Kong sub-sample, where over 40% of the firms have fully inde-pendent audit committees, but is not significant in Malaysia and Singapore where most firms have 2 out of 3 independent directors. This result indicates that the earnings–return relationship is stronger (i.e., earnings are more informative) when audit committees are fully independent, or monitoring is stronger when independent directors on the audit committee have an opportunity to talk outside of management's hearing. 5.2. Earnings informativeness and audit committee composition

InTable 5, we examine whether independence alone is effective or whether independent directors with expertise are associated with more reliable accounting information. Consistent with U.S. studies on financial expertise, the earnings–return relation does not improve with the proportion of independent non-accounting_{financial experts, but it does improve with the proportion of independent accounting} financial experts. The earnings–return relation is also positively related to the proportion of independent legal experts. Unlike the results for independence alone, the positive relationship is significant when a majority of audit committee members are independent and have either accountingfinancial or legal expertise. Moreover, the significance holds for both the Hong Kong and Malaysia/Singapore sub-samples (not reported). Thus, investors appear to associate the voluntary appointment of independent directors with greater accountingfinancial or legal expertise with more reliable reported earnings. Taken together, these results indicate, consistent with studies on board composition, that the composition of the audit committee is more important than its independence alone even though independence receives the most attention from policy makers. An alternate explanation is that investors view voluntary audit committee decisions, full independence and requirement of expertise, as more credible signals of the informativeness of earnings than compulsory majority independence.

5.3. Earnings informativeness, voting rights, and audit committee characteristics

InTable 6, we examine whether the increase in earnings informativeness associated with full in-dependence and a greater proportion of independentfinancial and/or legal experts is enough to offset the detrimental effect associated with concentrated control. We interact our measures of 100% audit committee independence and audit committee composition with voting rights and earnings to examine the relation between audit committee composition, conditional on level of voting rights. Similar to the findings inTable 5, wefind significant positive coefficients for the interactions of voting rights and earnings with both audit committees with 100% independence and with audit committees that include more independentfinancial or legal experts. In all cases, the coefficient for the audit committee interactions is larger than the coefficient for earnings conditional on voting rights alone. For example, the coefficient for the interaction of voting rights and earnings is−.52 (−.77) in specification 1 (3), but the coefficient when the 100% Audit Committee Independence (% Accountingfinancial experts) is added is 0.79 (1.14). In unreported tests, wefind no significant relation when audit committee independence is used in place of 100% independence. These results suggest that establishing a fully independent audit committee and appointing a majority of independent directors withfinancial or legal expertise are associated with more reliable reported earnings, and this certification more than offsets the detrimental effect on earnings informativeness that is associated with concentrated control.8 _{Our results also suggest that majority}

independence alone is not sufﬁcient to limit CEO inﬂuence and increase the perceived reliability of reported

8_{We examine what proportion of}_{firms that have 100% Audit Committee independence are also classified as Majority AF-L Expert} firms to see whether full independence is simply a proxy for a majority of independent directors with accounting financial or legal expertise. Only 29% offirms with fully independent audit committees are classified as Majority AF-L Expert firms, while 45% of firms with less than 100% audit committee independence are classified as Majority AF-L Expert firms. Moreover, both interactions remain significant when included in the same regression. Thus, full independence and majority expertise both appear to be important.

(14)

Table 5

Audit committee composition and earnings informativeness.

This table presents a regression analysis of the relation between audit committee composition and earnings informativeness as measured by the earnings–return relation.The dependent variable, CAR, is the cumulative 12-month market-adjusted stock return for the 12 months ending on the earliest report date for afirm's 2000 fiscal year earnings. Earnings equals reported net earnings for fiscal year 2000 divided by the market value of equity at the beginning of 2000. Voting Rights is the sum of the direct and indirect voting rights (a detailed description is provided inAppendix A). Cash Flow-to-Voting Rights is the ratio of the cashflow rights or ownership to the voting rights of the ultimate owner. %Non-accountingfinancial experts is the number of independent directors who are currently or were previously employed as executives in other publicly held corporations over the total number of directors on the audit committee. %Accountingfinancial experts is the number of independent directors with either experience as a Chief Financial Officer or a Certified Public Accountant over the total number of directors on the audit committee. %Legal experts is the number of independent directors who also practice law over the total number of directors on the audit committee. Majority AF-L expert equals one when more than 50% of the audit committee is represented by either accountingfinancial or legal experts. Size is the natural logarithm of the book value of assets in millions of U.S. dollars at the beginning of 2000. Market-to-Book is the market value of equity divided by the book value of total assets at the beginning of 2000. Leverage is the total liabilities divided by total assets at the beginning of 2000. All regressions include country and industryfixed effects (not reported). T-values are reported in parentheses. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.

Country and industryﬁxed effects Yes Yes Yes Yes

Intercept −37.53 (−2.06)⁎⁎ −41.32 (−2.25)⁎⁎ −35.17 (−1.92)⁎ −37.52 (−2.04)⁎⁎ Earnings 1.39 (2.61)⁎⁎⁎ 1.26 (2.38)⁎⁎ 1.06 (1.87)⁎ 0.97 (1.75)⁎ Earnings*Voting Rights −0.89 (−2.37)⁎⁎ −0.60(−1.54) −0.58(−1.5) −0.4(−1.02) Earnings*Cash Flow-to-Voting Rights 0.25

(0.83) 0.15 (0.49) 0.27 (0.91) 0.39 (1.27) Earnings*%Non-accountingﬁnancial experts 0.02

(0.07)

Earnings*%Accountingﬁnancial experts 0.46

(1.73)⁎

Earnings*%Legal experts 0.86

(1.69)⁎

Earnings*Majority AF-L expert 0.49

(2.38)⁎⁎ Earnings*Size −0.06 (−1.38) −0.06 (−1.56) −0.06 (−1.45) −0.07 (−1.68)⁎ Earnings*Leverage −0.01 (−0.03) −0.02(−0.06) 0.18 (0.52) 0.17 (0.51) Earnings*Market-to-Book 0.30 (1.98)⁎⁎ 0.30 (2.0)⁎⁎ 0.29 (1.9)⁎ 0.29 (1.94)⁎ Voting rights 0.07 (0.65) 0.05 (0.50) 0.05 (0.46) 0.03 (0.33)

Cash Flow-to-Voting Rights 4.10

(0.65) 3.75 (0.6) 3.22 (0.51) 1.67 (0.27) %Non-accountingﬁnancial experts −6.89

(−1.05)

%Accountingﬁnancial experts 4.08

(0.7)

%Legal experts −2.18

(−0.23)

Majority AF-L expert 2.08

(0.56) Size 1.30 (1.04) 1.36 (1.09) 1.14 (0.91) 1.29 (1.04) Leverage 2.77 (0.50) 3.42 (0.61) 1.32 (0.23) 2.28 (0.41) Market-to-Book −0.95 (−1.92)⁎ −0.97(−1.97)⁎⁎ −0.98(−1.99)⁎⁎ −0.94(−1.91)⁎ R2 0.22 0.23 0.22 0.23 N 450 450 450 450

(15)

Table 6

Audit committee characteristics, voting rights, and earnings informativeness.

This table presents a regression analysis of the relation between audit committee composition and voting rights on earnings informativeness as measured by the earnings–return relation. The dependent variable, CAR, is the cumulative 12-month market-adjusted stock return for the 12 months ending on the earliest report date for afirm's 2000 fiscal year earnings. Earnings equals reported net earnings forfiscal year 2000 divided by the market value of equity at the beginning of 2000. Voting Rights is the sum of the direct and indirect voting rights (a detailed description is provided inAppendix A). 100% Audit Committee independence equals one when all audit committee members are independent directors and zero, otherwise. %Non-accountingfinancial experts is the number of independent directors who are currently or were previously employed as executives in other publicly held corporations over the total number of directors on the audit committee. %Accountingfinancial experts is the number of independent directors with either experience as a Chief Financial Officer or a Certified Public Accountant over the total number of directors on the audit committee. %Legal experts is the number of independent directors who also practice law over the total number of directors on the audit committee. Majority AF-L expert equals one when more than 50% of the audit committee is represented by either accounting financial or legal experts. Size is the natural logarithm of the book value of assets in millions of U.S. dollars at the beginning of 2000. Market-to-Book is the market value of equity divided by the book value of total assets at the beginning of 2000. Leverage is the total liabilities divided by total assets at the beginning of 2000. All regressions include country and industryfixed effects (not reported). T-values are reported in parentheses. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.

Independent variable Dependent variable: CAR

Country and industryﬁxed effects Yes Yes Yes Yes Yes

Intercept −31.51 (−1.85)⁎ −32.57 (−1.89)⁎ −38.29 (−2.21)⁎⁎ −31.79 (−1.84)⁎ −34.95 (−2.01)⁎⁎ Earnings 0.87 (1.66)⁎ 1.15 (1.92)⁎ 1.46 (2.86)⁎⁎⁎ 1.01 (1.93)⁎ 0.91 (1.76)⁎ Earnings*Voting Rights −0.52 (−1.36) −0.77(−1.49) −0.77(−2.23)⁎⁎ −0.68(−1.89)⁎ −0.55(−1.53) Earnings*Voting Rights* 100% AC independence 0.79 (2.24)⁎⁎ Earnings*Voting Rights* %Non-accounting

ﬁnancial experts

−0.16 (−0.30) Earnings*Voting Rights *%Accounting

ﬁnancial experts

1.14 (2.16)⁎⁎

Earnings*Voting Rights *%Legal experts 1.91

(1.81)⁎

Earnings*Voting Rights *Majority AF-L expert 1.22

(2.76)⁎⁎⁎ Earnings*Size −0.03 (−1.08) −0.03 (−0.76) −0.06 (−1.8)⁎ −0.03 (−1.08) −0.03 (−1.05) Earnings*Leverage 0.07 (0.21) −0.09 (−0.27) −0.09(−0.28) 0.08 (0.23) 0.004 (0.01) Earnings*Market-to-Book 0.33 (2.30)⁎⁎ 0.35 (2.42)⁎⁎ 0.30 (2.12)⁎⁎ 0.32 (2.21)⁎⁎ 0.34 (2.43)⁎⁎ Voting Rights 0.04 (0.41) 0.07 (0.63) 0.05 (0.48) 0.05 (0.48) 0.02 (0.2) 100% Audit Committee Independence 7.08

(1.53)

%Non-accountingﬁnancial experts −6.14

(−0.94)

%Accountingﬁnancial experts 3.33

(0.58)

%Legal experts −3.21

(−0.33)

Majority AF-L expert 1.49

(0.4) Size 0.87 (0.71) 1.12 (0.9) 1.32 (1.07) 0.99 (0.81) 1.19 (0.97) Leverage 2.89 (0.52) 3.36 (0.06) 4.1 (0.74) 1.98 (0.35) 3.32 (0.6) Market-to-Book −0.95 (−1.94)⁎ −0.94(−1.91)⁎ −0.95(−1.93)⁎ −0.97(−1.97)⁎⁎ −0.93(−1.90)⁎ R2 0.23 0.22 0.23 0.22 0.23 N 450 450 450 450 450

(16)

earnings when control is concentrated. Moreover, independent directors who are executives at other companies are not associated with a stronger earnings–return relationship. The increase is found only for independent directors either with accountingﬁnancial or legal expertise, as they may possess greater ﬁnancial sophistication and/or objectivity.

5.4. Audit committee as substitute or complement

Our earlier results indicate that earnings informativeness is weakened when the ultimate owner of a firm has a higher level of control, but the divergence between control and cash flow rights is not significant. This may be because it is either difficult for investors to measure the divergence between cash flow and voting rights or because the protection of proprietary information persists during our sample period but the entrenchment effect from the divergence of cashflow to voting rights is mitigated by a stronger corporate governance environment during our sample period. Regardless, it is noteworthy that the earning-return relation may be weaker when proprietary information is protected, even when this private information is associated with higher returns. For example, the incentives of an ultimate owner with high cashflow rights may be aligned with other shareholders, and they may still choose to protect certain information if it is associated with the company receiving kickbacks (as long as the kickbacks are kept confidential). Thus, abnormal returns may be high for firms with high cash flow rights regardless of audit committee composition or earnings informativeness.

On the other hand, even though investors can observe high cashflow ownership, it may be difficult for other investors to differentiate between when controlling shareholders have an incentive to protect proprietary information and when the entrenchment effects of excess control outweigh the positive incentive effects of cashflow ownership. In this case, certain audit committee characteristics may certify the positive incentive effects of high cashflow ownership, and we might expect to find both stronger stock performance and a stronger earnings–return relation for firms with high cash flow ownership and certain audit committee characteristics.

5.4.1. Audit committee characteristics and market-adjusted returns

We begin by examining the CARs according to different levels of cashflow rights and audit committee characteristics. First, we divide the sample into High and Low cashflow rights sub-samples based on the median cashflow rights for the full sample. In Panel A ofTable 7, we further divide the High and Low cash flow rights sub-samples according to whether the audit committee is 100% independent or not. The results in Panel A suggest thatfirms with high cash flow rights and 100% audit committee independence have the

Table 7

Market-adjusted returns according to level of cashﬂow rights and audit committee characteristics.

CAR is the cumulative 12-month market-adjusted stock return for the 12 months ending on the earliest report date for afirm's 2000 fiscal year earnings. The High and Low cash flow rights sub-samples are created by dividing the sample according to the median cash flow rights for the full sample. In Panel A, we further divide the High and Low cash flow rights sub-samples according to whether the audit committee is 100% independent or not. In Panel B, we further divide the High and Low cashflow rights sub-samples according to whether more than 50% of the audit committee is represented by either accountingfinancial or legal experts or not.

Sub-sample CAR

Average

CAR Std. Deviation

Tukey's Studentized Range Test between each sub-sample and theﬁrst sub-sample Panel A: Cashﬂow rights and audit committee independence

High cashflow rights AND 100% Audit Committee independence 13.96 58.95 – High cashflow rights BUT NOT 100% Audit Committee independence 3.04 33.69 Insignificant Low cashflow rights AND 100% Audit Committee independence 1.90 40.26 Insignificant

Low cashflow rights BUT NOT 100% Audit Committee independence 1.44 29.73 Significant at the 10% level Panel B: Cashflow and audit committee profession

High cashﬂow rights AND Majority AF-L expert 21.26 60.38 –

High cashflow rights BUT NOT Majority AF-L expert 1.37 34.70 Significant at the 10% level Low cashflow rights AND Majority AF-L expert −5.46 27.02 Significant at the 10% level Low cashflow rights BUT NOT Majority AF-L expert 4.58 33.92 Significant at the 10% level

(17)

highest average 12-month market-adjusted returns, but the difference is only marginally significant when compared to the returns offirms with low cash flow rights and lower audit committee independence.

Panel B divides the High and Low cashflow rights sub-samples according to whether the majority of the audit committee directors are independent directors with accountingfinancial or legal expertise. In contrast to the results in Panel A, Panel B shows thatfirms with high cash flow rights and audit committees in which the majority of directors arefinancially or legally sophisticated independent directors have the highest average 12-month market-adjusted returns, and the returns are significantly greater than those of all of the other sub-samples. Thus,firms with high cash flow rights are associated with higher returns when the audit committee consists of a majority of independent directors withfinancial or legal expertise. Do thesefirms also have a stronger earnings–return relationship?

5.4.2. Audit committee composition, level of cashﬂow rights, and earnings informativeness

The results in Table 7 indicate that audit committee composition is significantly related to high abnormal returns when cashflow rights are higher. In this section, we examine whether audit committee composition is also signi_{ficantly related to increased earnings informativeness. We re-run the regression} analysis of audit committee composition and earnings informativeness separately for the High and Low cashflow rights sub-samples. The results for the High cash flow rights sub-sample are presented inTable 8. Similar to the results for the full sample, wefind significant, positive coefficients for earnings conditional on both fully independent audit committees and the proportion of independent directors with either accountingfinancial or legal expertise. No significant relation is found for independent non-accounting financial directors. In contrast, the results inTable 9indicate that audit committee characteristics are not significantly related to earnings informativeness for firms that are characterized by low cash flow rights.

Taken together, the results inTables 8 and 9support the hypothesis that audit committee charac-teristics are a complement to high cashflow ownership. Audit committee characteristics do not appear to provide a substitute mechanism for verifying the accuracy of reported earnings when incentives may be misaligned, that is, when cashflow rights are low. Instead, audit committee characteristics appear to provide a complementary mechanism for verifying the accuracy of reported earnings when earnings are viewed as being more opaque due to concentrated ownership and control. This suggests that controlling shareholders with high levels of cash flow rights can signal their commitment to a better corporate governance system by voluntarily establishing fully independent audit committees and audit committees with a majority of independent directors with accounting_{financial or legal expertise. This commitment,} which is currently voluntary in these countries, appears to be related to both stronger earnings infor-mativeness and stronger stock performance.

5.5. Robustness tests

In this section, we include other board and corporate governance characteristics that may be related to earnings informativeness to examine whether the relation between audit committee independence and composition is robust. For example,Xie et al. (2003)include additional audit committee characteristics (size and meeting frequency) and board characteristics (size, independence and whether the CEO serves as Chairman) in their study of earnings management andfind that audit committee meeting frequency and board size are negatively related to discretionary current accruals.Becker et al. (1998)andFrancis et al. (1999)find that earnings quality is positively related to audit quality, or being audited by a Big 6 Auditor, during their sample period. We therefore collect these variables for our sample. In untabulated results, we find that boards in our sample tend to meet the minimal board independence requirements during this period, and thus, primarily appoint independent directors to serve on the audit committee.9Only 36% of all directors are independent, on average. We alsofind little variation in audit quality. Over 90% of reporting firms hire a Big 5 Auditor.10_{Finally, only 272 (229)}_{firms report the number of times the audit committee}

9

Firms primarily appoint independent directors with expertise to serve on the audit committee in our sample. Thus, we do not include a separate board measure for independent director expertise.

10